How Useful Are Banks' Earnings-A-Risk And Economic Value Of Equiy-A-Risk Public Disclosures? Jacques Préfonaine, (Email: jprefonaine@adm.usherbrooke.ca), Universié de Sherbrooke, Canada Jean Desrochers, (Email: jdesrochers@adm.usherbrooke.ca), Universié de Sherbrooke, Canada ABSTRACT This paper examines he informaion conen and he usefulness of banks' ineres rae risk public disclosures. ALM managers use Earnings a Risk ( EAR ) and Economic Value of Equiy a Risk ( EVEAR ) as measures of he dollar amoun of poenial loss o ne ineres income and common shareholders' equiy as a resul of unforeseen ineres rae changes. These wo ineres rae risk managemen merics are now recognized benchmarks for measuring ineres rae risk exposure, and is poenial impac on a bank's financial posiion. A he explici reques of regulaors, financial analyss and compeiive pressures, more commercial banks are now reporing EAR and EVEAR numbers in heir annual financial repors. To examine preliminary evidence on he informaion conen of such public disclosures, we composed a sample of some of Norh America's larges commercial banks. The Canadian peer group is based on Canada's seven larges banks, and he U.S. peer group is composed of welve of is larges banks. In paricular, we invesigae if "ex ane" EAR and EVEAR numbers help regulaors, financial analyss and invesors o explain he subsequen variabiliy of commercial banks' ne ineres income and ne income over ime. INTRODUCTION A LM managers use Earnings a Risk (EAR) and Economic Value of Equiy a Risk (EVEAR) as measures of he dollar amoun of poenial loss o ne ineres income and common shareholder's equiy as a resul of unforeseen ineres rae changes. These wo ineres rae risk managermen merics are now recognized benchmarks for measuring he non-rading exposure o ineres rae risk, and is poenial impac on a bank's financial posiion. Similarly, he Value a Risk (VAR) measure is he recognized benchmark for measuring he rading exposure o all marke risks, including ineres rae risk, and is poenial impac on a bank's financial posiion. A he explici reques of regulaors, financial analyss and compeiive pressures, more inernaionally acive commercial banks are now reporing EAR and EVEAR numbers in heir annual financial repors. Lopez (2003) explains he ongoing inernaional effors o improve he regulaion and supervision of banking insiuions o reflec advances in financial risk managemen echniques. His analysis suppors he view ha improved public disclosures regarding heir condiions, operaions and risk managemen informaion lead o increased ransparency and should lead o more effecive marke discipline. The evoluion of bank disclosure sandards in he Unied Saes has been described in a sudy published by he BGFRS (2000). The SEC and he FASB ogeher se he core disclosure requiremens for publicly raded banks. Moreover, all banks in he U.S. are required o file quarerly regulaory repors. In a recen publicaion, he Basel Commiee on Banking Supervision (BCBS, 2003) provides an overview of he disclosure pracices of a sample of inernaionally acive commercial banks. The BCBS survey focuses on he annual repors of 54 banks headquarered in he commiee's member counies. I included 104 quesions addressing quaniaive and qualiaive disclosures in 4 various caegories: capial adequacy, marke risk inernal modeling, 87
derivaives, accouning and presenaion policies The survey reveals ha many banks have coninued o expand he exen of heir disclosures. The main findings of he disclosure survey ha relae more direcly o he focus of his sudy are he following: Firsly, disclosure of informaion on inernal risk models was much more common for marke risk han for credi risk. Secondly, he mos noeworhy improvemen is he increase in he disclosure of informaion on oher risks (operaional and legal risks, liquidiy risk and ineres rae risk in he banking book (non-raded). Thirdly, regarding individual disclosure iems, he survey resuls indicae ha marke risk inernal modeling (e.g. he ype used) was one of he mos common iems disclosed. The res of he paper is srucured as follows: secion wo defines he EAR and EVEAR non-rading exposure measures; he VAR rading exposure measure is also examined as well as is predicive power. The hird secion describes he research mehods, he bank sample daa, and formulaes hypoheses on he usefulness of EAR and EVEAR numbers o explain he subsequen variabiliy of commercial banks' earnings and economic value of equiy. Secion four presens and discusses he sudy's empirical resuls. Finally, he conclusion, limis of he sudy and suggesions for furher research are drawn in he fifh secion. IMPROVED MARKET RISK INFORMATION DISCLOSURES Recenly, several auhors have examined if improved marke risk informaion disclosures lead o increased ransparency and more effecive marke discipline. Wong (2000) sudied he associaion beween SFAS no. 119 derivaives disclosures and he foreign exchange risk exposure of manufacuring firms. Chrisofferson, Hahn and Inoue (2001) esed, compared and combined value a risk measures. Berkowiz and O'Brien (2001) evaluaed he accuracy of VAR a commercial banks. In anoher sudy, Linsmeier, Thornon, Venkaachalan and Welken (2002) analyzed he effec of mandaed marke risk disclosure on rading volume sensiiviy o ineres rae, exchange rae and commodiy price movemens. Of more direc imporance o he focus of his sudy, Lopez (2003) summarizes he conclusion of a case sudy repored by he BGFR (2000) regarding SEC requiremens for disclosure of marke risk exposures. The auhor defines marke risk exposures as poenial financial losses due o adverse movemen in securiies marke prices. Mos ofen, commercial banks repor such risks wih value-a-risk (VAR) esimaes ha summarize he poenial losses ha migh occur wih a specified probabiliy (95 % or 99 % of he ime) over a given ime horizon like one or 10 rading days. A bank disclosing, for example, ha is daily VAR is $25 million a he 99 %level, indicaes ha here is only a 1 % chance he bank will incur more han a $25 million rading loss over he nex day. "In he case sudy, bank VAR disclosures were found o vary in deail across banks and o have an unclear connecion wih acual rading performance during he urbulen hird quarer of 1998". The auhor finds ha even hough such heerogeneiy is presen in hese ypes of public disclosures, he academic lieraure sill suggess ha marke paricipans can assess bank risks accuraely. In anoher more recen sudy of VAR disclosures, Jorion (2003) found ha VAR numbers in quarerly and annual repors, from 1995 o 2000, of 8 publicly raded U.S. commercial banks provided reasonable predicions of he subsequen variabiliy of heir rading revenues. Thus, he empirical resuls presened in he Jorion sudy sugges ha VAR disclosures are informaive in ha hey predic he variabiliy of rading revenues. Thus, analyss and invesors can use VAR disclosures o compare he risk profiles of banks' rading porfolios. Figure 1 shows an example of Sun Trus Banks' EAR and EVEAR disclosures. Noice ha his bank uses Ne Ineres Income-a-Risk insead of he more common EAR measure; neverheless, boh measures are compaible. In addiion, managemen's discussion provides valuable disclosures ino he ineres risk modeling process a Sun Trus Banks. 88
FIGURE 1: Excerps from he Annual Repor of Sun Trus Banks, 2002, pp. 37-38, Managemen's Discussion of Marke Risk from Non-Trading Aciviies. 89
TESTING THE INFORMATION CONTENT OF EARNINGS-AT-RISK AND ECONOMIC VALUE OF EQUITY-AT-RISK PUBLIC DISCLOSURES EAR The EARNINGS-AT-RISK measure represens an ex ane esimae of changes in earnings over he nex welve monhs should ineres rae change by + or 100 basis poins. This formulaion of shor-erm ineres sensiiviy analysis is performed and disclosed by mos financial insiuions and faciliaes comparisons beween peers. EVEAR The ECONOMIC VALUE OF EQUITY-AT-RISK measure represens an ex ane esimae of ne change beween he presen value of asses and he presen value of liabiliies should ineres rae change by +/- 100 basis poins. This formulaion of longer-erm ineres sensiiviy analysis is also performed and disclosed by many financial insiuions and furher faciliaes comparisons beween peers. Proposed Tesing Procedure We posulae ha hose insiuions wih he lowes (highes) ex ane relaive EAR measures should display he lowes (highes) ex pos relaive changes in heir earnings as a resul of a given change in ineres rae levels. To do so, he following equaion is esimaed : Δ E E + 1 = a + b EAR E (1) Where E +1 measures he dollar change in ne ineres income from period o period +1, E measures he period ne ineres income in dollars, EAR represens he shor-erm ineres rae risk dollar disclosure a period. Expressing boh he l.h.s. and r.h.s. of equaion (1) in relaive values allows comparisons o be carried ou over ime wih he same insiuion, and beween insiuions which have differen earnings. We propose o es equaion (1) empirically using he OLS regression mehod. As shown furher on, our daa sample consiss of weny Norh American commercial banks. Here, we posulae ha hose insiuions wih he lowes rank (highes rank) in heir ex ane relaive EAR measures should also display he lowes rank (highes rank) in heir ex pos relaive changes in earnings as a resul of a given change in ineres rae levels. To do so, he following equaion is esimaed: Rank (i) Δ E E EAR +1 = a + b Rank (i) E (2) Where Rank (i) varies from one (lowes ex ane relaive EAR measure or lowes relaive change in earnings) o weny (highes ex ane relaive EAR measure or highes relaive change in earnings). Expressing boh he l.h.s. and r.h.s. of equaion (2) in rank values allows more general comparisons o be carried ou over ime beween insiuions. We propose o es equaion (2) empirically using he Spearman rank correlaion coefficien. Since he procedure for esing he informaion conen of EAR and EVEAR public disclosure measures are pracically similar, hey are no repeaed here for EVEAR. Sample Descripion To examine preliminary evidence on he informaion conen of banks' EAR and EVEAR public disclosures. I composed a sample of some of Norh America's larges commercial banks. The Canadian bank peer group is based 90
on Canada's seven larges domesic banks, and he U.S. peer group is composed of hireen of is larges commercial banks. Table 1 shows for each bank in he daa sample is icker symbol, oal asses expressed in U.S. dollars, senior long-erm deb credi raings, EAR and EVEAR annual daa availabiliy. TABLE 1: DATA SAMPLE DESCRIPTION 1 2 S&P S.L.T.D. Raings 2 EAR Year Available EVEAR Year Available Bank Ticker Symbol Toal Asses U.S. ($ Billions) 1 1. Bank of Monreal BMO $ 161,5 AA- 1995 1995 2. CIBC CM $ 174,5 A+ 1998 1998 3. Laurenian BoC. LB $ 11,9 A- 2000 2000 4. Naional BoC. NA $ 47,6 A 1999 1999 5. Royal BoC. RY $ 243,9 AA- 1996 1996 6. Scoiabank BNS $ 189,2 A+ 1996 1996 7. TD Bank TD $ 177,5 A+ 1995 1995 8. Bank of America Corp. BAC $ 660,5 A+ 1999 No disclosed 9. Bank One Corp. ONE $ 277,4 A 1997 +/- disclosed 10. Ciigroup C $1 097,2 AA 1996 1998 11. Flee Boson F. Corp. FBF $ 190,5 A 1994 1996 12. JP Morgan Chase JPM $ 758,8 AA- 1993 +/- disclosed 13. Key Corp. KEY $ 85,2 A 1994 +/- disclosed 14. Mellon F. Corp. MEL $ 36,2 A+ 1994 No disclosed 15. Naional Ciy Corp. NCC $ 118,3 A 1993 1993 16. PNC F.S. Group PNC $ 66,4 A- 1997 1997 17. Sun Trus Banks STI $ 117,3 AA- 2000 1999 18. US Bancorp. USB $ 180,0 A 1997 +/- disclosed 19. Wachovia Corp. WB $ 341,8 A 1993 No disclosed 20. Wells Fargod Cy. WFC $ 349,3 A+ 1994 No disclosed Canadian banks' oal asses as of Ocober 31, 2002 were ranslaed a 0,6385 x Canadian $ ino U.S. funds. Moody's long-erm deb credi raings were also available. TABLE 2: EAR RESULTS - CANADIAN BANKS Variables Value F Tes T Tes Sig. Regression R.Sq. 4.528 0.041 12.1 % a.0735 1.430 0.162 b (sd.) -0.3470-2.128 0.041 As can be gahered from Table 1, he average size of he seven Canadian banks a $143 B is no ha differen from he hireen U.S. a $329.1 B if we remove he hree larges U.S. banks (BAC, C and JPM). I is also possible o compue he average defaul risk of he seven Canadian banks and o compare i o he average defaul risk of he hireen U.S. banks. In order o carry ou his analysis a bank's senior long-erm deb credi raing is simply replaced by a number. For example, AAA equals 1, AA+ equals 2, AA equals 3, AA- equals 4, A+ equals 5, A equals 6, and A- equals 7. The Canadian bank sample has an average credi raing of 4,79; ha is, a shade lower han he U.S. bank sample a 4,69. Also imporan is he fac ha all weny Norh American banks in our sample make EAR disclosures in heir annual repors. All seven Canadian banks also make EVEAR disclosures in heir annual repors. In he U.S. bank sample, four banks disclose EVEAR numbers, one oher bank, Ciigroup, discloses an equivalen number using anoher measure, four banks make sporadic disclosures, hree banks do no disclose EVEAR numbers, and finally one bank (Mellon F. Corp.) carries ou EVEAR compuaions bu hey are no disclosed in is annual repor. Since EAR disclosures are more complee and frequen in our daa sample han EVEAR disclosures, i sands o reason ha more analysis can be carried wih EAR disclosures in his sudy. 91
THE EMPIRICAL RESULTS The objecive of he empirical ess is o examine if commercial banks' ex ane EAR numbers are relaed o ex pos variaions in ne ineres income for he complee bank sample, he U.S. bank sub-sample, and he Canadian bank sub-sample during he sudy period. Similar ess were also carried ou o examine if commercial banks' EVEAR numbers were relaed o ex pos variaions in ne ineres income for he Canadian and U.S. sub-samples, and also he complee commercial bank sample. OLS regression resuls indicae ha ex ane EAR numbers were no closely relaed o ex pos variaions in ne ineres income for he complee bank sample and he U.S. bank sub-sample. As shown in Table 2 below, he resuls were only saisically significan for he Canadian bank sub-sample ( a he 96 % level ). In addiion, OLS resuls also lead us o conclude ha ex ane EVAR numbers were no closely relaed o ex pos variaions in ne ineres income for he complee bank sample, nor for he U.S. and Canadian bank sub-samples during he sudy period. Given he general one of he resuls presened above on ex ane EAR and EVEAR numbers and subsequenly observed variaions in commercial banks' ne ineres income, he Spearman rank correlaion coefficien analysis which was proposed earlier was no carried ou. CONCLUSIONS The sudy documens ha EAR public disclosures were made in heir annual financial repors by all commercial banks in our sample. In comparison, EVEAR public disclosures were made by all Canadian banks, bu by only a fracion of U.S. banks in our sample. ALM managers use EAR as a shor-erm measure, and EVEAR as a longer-erm measure of he dollar amoun of poenial loss o ne ineres income and common shareholders' equiy as a resul of unforeseen ineres rae changes. These wo ineres rae risk managemen merics are now recognized benchmarks for measuring he exposure o ineres rae risk, and is poenial impac on a bank's financial posiion. A he explici reques of regulaors, financial analyss and compeiive pressures, more commercial banks are now reporing EAR and EVEAR numbers in heir annual financial repors. In addiion, some banks like JPMorgan Chase in he U.S. have been reporing non-rading porfolio VaR numbers for invesmen porfolio and A/L aciviies. To examine preliminary evidence on he informaion conen of such EAR and EVEAR public disclosures, we composed a sample of some of Norh America's larges commercial banks. The Canadian bank sub-sample is based on Canada's seven larges domesic banks, and he U.S. sub-sample is composed of welve of is larges commercial banks. In paricular, we invesigaed if ex ane EAR and EVEAR numbers help regulaors, financial analyss and invesors o explain he subsequen variabiliy of commercial banks' ne ineres income and ne income over ime. Unlike in Jorion's (2003) sudy of commercial banks' VaR public disclosures, he preliminary resuls presened in his paper indicaed ha banks' EAR and EVEAR public disclosures did no explain he subsequen variabiliy of heir ne ineres income over ime. Canadian banks' EAR public disclosures were found o provide some indicaion of he subsequen variabiliy of heir ne ineres income during he period under sudy. Daa availabiliy: The daa used in his sudy can be obained from public sources. 92
REFERENCES 1. Basel Commiee on Banking Supervision ( BCBS ). 2004. Principles for he Managemen and Supervision of Ineres Rae Risk. Basel, Swizerland: BIS. 2. Basel Commiee on Banking Supervision ( BCBS ). 2003. Public Disclosures by Banks: Resuls Of he 2001 Disclosure Survey. Basel, Swizerland: BIS. 3. Basel Commiee on Bank Supervision. 2003a. Overview of he New Basel Capial Accord: Consulaive Documen (April). 4. Basel Commiee on Bank Supervision. 2003b. Public Disclosures by Banks: Resuls of he 2001 Disclosure Survey. Basel Commiee Publicaions No. 97 (May). 5. Berkowiz, J. and J. O'Brien. 2001. How accurae are he value a risk models a commercial banks? Working paper, Board of Governors of he Federal Reserve Sysem. 6. Board of Governors of he Federal Reserve Sysem. 2000. Improving Public Disclosure in Banking. Saff Sudy # 173. 7. Board of Governors of he Federal Reserve Sysem. 2001. Supervisory Leer 01-06: Enhancemens o Public Disclosure. 8. Chrisofferson, P., J. Hahn, and A. Inoue. 2001. Tesing, comparing, and combining value a risk measures. Journal of Empirical Finance 8 (July): 325-342. 9. Jorion, P. 2002. How Informaive Are Value-a-Risk Disclosures? The Accouning Review 77 ( Ocober ): 911-931. 10. Jorion, P. 2003. How Informaive Are Value-a-Risk Disclosures? The Accouning Review 77, pp. 911-931. 11. Kwan, Simon H. 2002. The Promise and Limis of Marke Discipline in Banking. FRBSF Economic Leer 2002-36 (December 13). 12. Kwan, Simon H. 2002. Bank Securiy Prices and Marke Discipline. FRBSF Economic Leer 2002-37 (December 20). 13. Linsmeier, T., D. Thornon, M. Venkaachalam, and M. Welker. 2002. The effec of mandaed marke risk disclosure on rading volume sensiiviy o ineres rae, exchange rae, and commodiy price movemens. The Accouning Review 77 (April): 343-378. 14. Lopez, Jose A. 2003. How Financial Firms Manage Risk. FRBSF Economic Leer 2003-03 (February 14). 15. Lopez, Jose A. 2003. Disclosure as a Supervisory Tool: Pillar 3 of Basel II. FRBSF Economic Leer 2003-22 (Augus 1). 16. Wong, F. 2000. The associaion beween SFAS No. 119 derivaives disclosures and he foreign exchange risk exposure of manufacuring firms. Journal of Accouning Research 38 (Auumn): 387-417. 93
NOTES 94